EC50 images, a novel endpoint from PET target occupancy studies, reveal spatial variation in apparent drug affinity

Abstract

We recently introduced voxel-level images of drug occupancy from PET via our “Lassen plot filter.” Occupancy images revealed clear dependence of 11C-flumazenil displacement on dose of GABAa inhibitor, CVL-865, but with different scales in different brain regions. We hypothesized that regions requiring higher drug concentrations to achieve desired occupancy would have higher EC50 values. We introduce an “EC50 image” from human data to evaluate this hypothesis. Five healthy subjects were scanned with the nonselective GABAa tracer, 11C-flumazenil, before and (twice) after administration of CVL-865. We created ten occupancy images and applied an Emax model locally to create one EC50 image. We also performed simulations to confirm our observations of regional variation in EC50 and to identify the main source of variability in EC50. As expected, the EC50 image revealed spatial variation in apparent drug affinity. High EC50 was found in areas of low occupancy for a given drug dose. Simulations demonstrated that sampling from an inadequate range of plasma drug concentrations could impair precision. Our results argue for (a) confidence in the ability of the EC50 images to identify regional differences and (b) a need to tailor the range of drug doses in an occupancy study to regularize the precision of the EC50 throughout the brain. The EC50 image could add value to early-phase drug development by identifying regional variation in affinity that might impact therapy or safety and by guiding dose selection for later-phase trials.